Incorporate speckle reduction algorithms?

[csheaff]

Hi, I just discovered ITKUltrasound. As you guys probably know, reducing speckle can have a significant impact on image segmentation. Have you thought about incorporating a suite of different algorithms for doing this? Might want to take a look at this MATLAB toolbox and a paper written about it:

https://www.mathworks.com/matlabcentral/fileexchange/54044-image-despeckle-filtering-toolbox?focused=5709994&tab=function

http://www.sciencedirect.com/science/article/pii/S0169260714000327

[thewtex]

@csheaff Yes, we intend to add despeckle methods. Thank you for the link and suggestion. It looks like a good open source / open science algorithm. We will add it here, and our implementation should be real-time. I will also add some additional despeckling algorithms. Any other addition suggestions or contributions are welcome.

[csheaff]

great. i might also suggest adding Oriented Speckle Reducing Anisotropic Diffusion (OSRAD) by Krissian. From what i've read from review papers, it offers the best balance between performance and efficacy. There's a C++ implementation of his somewhere on the web that is free for download, but I can't seem to find it.

Edit: It used to be located here, but no longer:
[http://serdis.dis.ulpgc.es/~krissian/HomePage/Demos/OSRAD/OSRAD.html]

Fortunately I downloaded it before it was taken down. I think it uses ITK, incidentally. It is attached.

OSRAD.zip

[thewtex]

@csheaff Excellent, thanks!

[agirault]

As discussed with @tbirdso and @vicory, the OSRAD source code is LGPL and cannot be integrated into ITK as-is, so it might be reimplemented in ITK to be compatible with Apache 2.0.

@karlkrissian we saw you had contributed to ITK/VTK/Slicer in the past which was very exciting to us! Let us know if you'd be interested in helping port the OSRAD algorithm to ITK: even just getting your guidance & reviews would be extremely beneficial.

[karlkrissian]

Hello, I will probably be able to help during my free time.

[agirault]

Merci @karlkrissian!

[karlkrissian]

It is ok for me to relicense the code from LGPL to Apache 2, before doing it, I will have to check two files contributed by another author: CurvaturasPrincipales.*pp.